Convert PDF to Markdown
Turn a PDF into structured Markdown text: headings, paragraphs, lists, tables and links are rebuilt, and page headers and page numbers removed. Ideal for giving a document to ChatGPT, Claude or your wiki.
- Free
- Headings, lists, tables
- Token counter
- No upload
100% private — Everything is processed directly in your browser: your files, voice, and image are never sent to Chatzam's servers.
How to convert a PDF to Markdown
-
Upload the PDF
Drop the file in: its pages are read right in your browser.
-
Adjust the conversion
Keep the whole document or just a few pages, and choose whether to remove headers, mark pages or flag images.
-
Copy or download
Check the result as raw Markdown or as a preview, then copy it or download the .md file.
Markdown true to the PDF's structure
Headings and formatting
Font sizes become #, ## and ### headings, and bold and italics are kept.
Columns and paragraphs
Two-column layouts are read in the right order, lines are joined back into paragraphs and words hyphenated at line ends are rejoined.
Lists, tables and links
Bullets, numbered lists, simple tables and clickable links are converted to Markdown syntax.
Built for AI
Repeated headers, footers and page numbers are removed, and a counter estimates the number of tokens.
Why convert a PDF to Markdown?
Markdown is plain text with a few symbols added: a hash for a heading, a dash for a bullet, vertical bars for a table. It's readable without any software, editable in any editor and displays cleanly on GitHub, Notion, Obsidian or a company wiki. It's also the format AI assistants understand best: when you paste a report into ChatGPT or Claude, the heading hierarchy and tables help it summarize, compare or answer a specific question. A plain copy and paste from a PDF reader often mixes up columns, breaks sentences at every line and repeats page headers; converting to Markdown avoids those problems.
How is the structure rebuilt?
A PDF contains no headings or paragraphs, only pieces of text placed at exact coordinates. The tool identifies the body text's font size, then ranks the larger sizes: the largest becomes a level 1 heading, the next a level 2, and so on. It detects columns from the empty space between them, joins the lines of a paragraph back together and removes hyphenation. Lines aligned into several cells form a table, and bullets and numbers start a list. Text repeated at the top or bottom of most pages, such as a report title or “Page 3 of 12”, is dropped.
Tokens, limits and scanned PDFs
AI models measure text length in tokens, which are chunks of words. The counter gives a ballpark figure, about one token per four characters: handy for knowing whether a document fits in a conversation or whether you'd better send only some pages. Images aren't converted, but their position can be flagged. Very graphic layouts, like a brochure or a table with merged cells, sometimes need a quick review. Finally, a scanned PDF only contains images of pages: the tool tells you so and suggests running OCR first to make it readable. The file is processed in your browser and never uploaded.
Frequently Asked Questions
How do I give a PDF to ChatGPT without losing the formatting?
Convert it to Markdown: headings, lists and tables stay readable for the AI. Paste the result into the chat or attach the .md file.
Are the PDF's tables preserved?
Simple tables with well-aligned columns become Markdown tables. Tables with merged cells or heavy graphic design may need touching up.
Is my PDF uploaded to a server?
No. The PDF is read and converted entirely in your browser; it never leaves your device.
Why is the result empty?
Your PDF is probably a scan: its pages are images with no text. First make it searchable with the “OCR PDF” tool, then convert it.
What does the token count mean?
It's an estimate of the text's length as AI models count it, based on roughly four characters per token. The exact number depends on the model used.